492 METALLURGY OF CAST IRON. bars cast flat stood the greatest load, with their side which was down when cast being in extension when tested, and also that the greatest difference in this re-spect exists in the round bar. Again I would call at-tention to the fact that the results in all the flat cast jbars were very erratic. This Table compares very closely in averages with a large number of tests which I have made on this point to satisfy myself as to the correctness of such results, and they always point in one direction. A deceptive point which it might be well to notice in casting test bars flat is the chance it affords of making a test bar record too great a strength for an iron. Take a round bar cast flat and test it with its side cast down in extension, or as illustrated in Fig. 109, page 490, and one can record a greater strength than by any other method of casting; but where one desires to record the honest and natural strength of an iron, he should use the round bar cast on end. And by a comparison of the round bar cast on end with those cast flat, as seen by Tables 102 and 103, next page, the system which the author advocates is found to be one which will not permit a tester to obtain a greater strength than that which the iron truly possesses, nor admit of any jugglery in recording tests. When it is known that one side of a flat cast bar can often give 300 to 400 pounds more strength than its opposite side, there is surely an opening for deception and variable results. The mixture of iron charged for the test on next page was all pig metal of the analysis seen in Table 104. The analysis of the test bars shows the silicon to be reduced ten points and the sulphur doubled by re-melting the iron.rally denser, or of a closer gram, than that composing' the upper half of the Chapter I,XX,, par,*-*; :;y.| in r,.s.j. and attacked the usual formulae for loaded beams asound burs..........................*fi$7 "